Last month, Zycus held its annual user conference, Horizon 2026 in Beaver Creek, Colorado, bringing together hundreds of customers, partners, and sponsors for the two day event. Finely curated (as always) and set against the Rocky Mountains, the conference offered wonderful vistas plus an expansive view of Zycus’ focus on and approach to Agentic AI, including a major platform announcement (Merlin MAX).
All attendees left with cowboy hats (my first!), which served as props for an onstage analyst “showdown” where I was joined by Patrick Reymann of IDC and Chris Sawchuk of The Hackett Group. The three of us debated some of the big questions about the future of procurement, including autonomous sourcing, supplier onboarding, category management, tail-spend negotiation, and contract value leakage, all in the context of agentic AI.
The themes of the showdown were consistent with the event focus on what comes next for Source-to-Pay and how quickly AI can be used in “production,” to augment procurement operations.
Beyond S2P: The CEO Keynote
As he does each year, Zycus’ CEO Aatish Dedhia set the direction of the conference in his keynote, “Beyond S2P: Autonomous Procurement from Intake to Outcomes.” He described a clear change in customer conversations around AI over the past year; his team is seeing what it calls increasing “AI hunger,” with procurement leaders showing they are more than just AI-curious, asking harder and more direct questions about deployment, scale, governance, and business value. Aatish tied that shift to his company’s longer history in AI.
The centerpiece of his presentation was the announcement of Merlin MAX, Zycus’ new platform for autonomous AI execution that will arrive in mid-2027. Today, Zycus’ “Agentic Flows” operate on top of the current Zycus S2P platform. Merlin MAX introduces a new AI-first S2P suite built around what Aatish calls three primary towers: (1) the Intake tower which serves as the enterprise’s front door to procurement; (2)the Strategy tower which supports human-led planning and decision-making; and (3) the Outcomes & Autonomy tower is where agentic execution operates.
Aatish explained that Zycus will continue to innovate on its current S2P suite with new agentic capabilities while developing the new platform and that their convergence should be seen as a progressive evolution of the current offering, rather than a big bang replacement. This suggests that Zycus views the path for its current customers to move to Merlin MAX will look more like an upgrade than a migration. We hope to hear more details soon regarding the new solution.
With Merlin MAX, Aatish says AI will become the new UI and that Agentic flows will become the new applications. This means that a business request will be able to trigger sourcing, autonomous negotiation, contract review, supplier onboarding, renewals, or value-leakage activity without requiring the user to understand which traditional/underlying application should handle it.
Beneath Merlin MAX’ agentic layer sits the Merlin Trusted Core, which Zycus defined as the guardrails for controlled autonomy. This core will bring together governance, data, access, responsibilities, integrations, and enterprise controls to manage what agents can access, decide and execute. This governance layer will allow procurement teams to evolve their AI governance as AI capabilities extend into their operations.
Zycus was an early ProcureTech mover into AI and this announcement is very much in-line with the company’s history and two decades+ of innovation. It is also a strategy that more and more ProcureTech companies are making (whether announced formally or not).
The Mainstage Demo: From Agentic Flows to Autonomous Outcomes
With Merlin MAX on the roadmap, Zycus’ current solutions and strategy were on full display including many agentic workflows demonstrated throughout the event. For example, Merlin Intake takes a basic business request, classifies it, asks for missing information, identifies potential risk, and routes the request into the appropriate workflow. Zycus described the interaction as “one line in, five critical questions out,” illustrating how an intelligent intake layer can reduce the amount of procurement process knowledge required from business users.
For tactical and tail spend, Zycus pushes the model further with ANA, its Autonomous Negotiation Agent. ANA is designed to engage suppliers directly and conduct multi-round negotiations based on objectives and parameters established for the event. In the demonstration, ANA negotiated unit price, but also, payment terms, lead times, and standard SLAs. Savings can be prioritized in one event, while speed or other criteria can carry greater weight in another.
Zycus says that its ANA solution uses sentiment analysis during supplier interactions to help determine when continuing to press a negotiation may become counterproductive. More broadly, ANA agents are designed to evaluate the supplier’s offer against the objectives established by the team.
One of the more interesting new use cases was Contract Value Leakage. The Zycus agent reads contracts, calculates contractual entitlements, compares those terms against actual transactions, and identifies missed rebates, volume discounts, pricing discrepancies, credits, and other sources of leakage. It can continue working until an issue reaches a point where human judgment is required. Zycus also previewed autonomous supplier onboarding and a new Category Management offering focused on AI-powered intelligence, scenario analysis, and guided decisions.
Customers Put the Vision in Context
Several customer presentations helped ground the technology discussion. Ben Cutting of TKC Holdings described the challenge of creating a more unified procurement model across businesses spanning correctional foodservice, commissary and technology operations, and hospitality supplies. Procurement must coordinate closely with legal, IT, and finance, making the transformation an orchestration challenge as much as a technology project.
TKC’s evaluation process was particularly interesting. The team began with a broad provider field and ultimately reached a shortlisted group. ne slide summarized its philosophy well: “We didn’t evaluate slide decks. We evaluated thinking under pressure.” TKC ultimately selected Zycus, citing its CLM capabilities and the company’s broader vision. Although its roadmap extends from intake through sourcing, contracts, suppliers, procurement, and ultimately outcomes, TKC reported early, strong sourcing savings results.
Meanwhile, Ann-Marie McCarthy, Head of Global Procurement at Paul Hastings LLP and a former CPO Honors finalist, described a very different starting point for her firm. The $2.7 billion multinational law firm had no integrated S2P platform, contract approvals could average six months, supplier onboarding could take months, and highly compensated professionals were spending time on low-level AP and procurement activities.
After evaluating six major S2P providers, McCarthy and her team selected Zycus, highlighting Merlin Intake inside Microsoft Teams, Zycus’ understanding of the firm’s requirements, and its willingness to co-innovate around legal-specific needs.
The Analyst View
Without a doubt, Horizon 2026 reinforced Zycus’ direction and focus. The company has been moving beyond adding AI functionality to the traditional S2P suite to deliver a solution where agents are an increasingly important part of how procurement work gets done. With the launch of Merlin MAX in 2027, Zycus plans to fully merge that strategy with its solutions.
Of course, there is always a distinction between a solution provider’s vision and what is broadly available today. But many of the capabilities shown at Horizon are available now and being adopted by early customers.
Procurement teams have spent decades deploying solutions that drive value. Agentic AI creates the opportunity for technology to expand procurement’s reach and impact. And while the evaluation of agents in a ProcureTech RFP may be quite difficult today, it is an area worthy of focus. Attendees at Horizon 2026 left with valuable education, practical examples, and a clearer view of what to look for as they evaluate agentic capabilities.
